Minimizing Sampling in Google Analytics Reports

The Google Analytics connector includes a new capability to minimize sampling problems with large sites and long time frame queries. Check the box on the Sort tab, and Analytics Edge will make separate queries for each period in the selected date range and merge the results into a single table. To be effective, it must be used with a Date, Week, or Month time dimension.



Share on LinkedInTweet about this on TwitterShare on FacebookPin on PinterestShare on Google+Email this to someone

Analytics Edge will automatically make a separate query for each time period in the query and merge the results. Since each query covers a shorter date range, the sampling error is reduced if not eliminated.

Mixed time dimensions, accurate data

Note that if you use multiple time dimensions in the query, like Date and Month of Year, Analytics Edge will make queries for the smallest — Date in this example.  You will always get the finest resolution possible, and the most accurate numbers; Analytics Edge always queries Google Analytics asking for the HIGHER_PRECISION sampling level.

Combining results for longer time periods

If you want to get measures for longer time periods, such as monthly, but you need to query daily to reduce sampling to an acceptable level, you should not use rate or average metrics in your query, Mathematically calculating the average of daily averages for the month will produce the wrong number. You should include the root numbers for the calculated metrics (like Bounces and Sessions for Bounce Rate), add them up for the month, then perform the calculation (Bounce Rate = Sum of Bounces / Sum of Sessions).

Analytics Edge makes this process easy; for example:

  • query dimensions Month of Year and Date, plus metrics Bounces and Sessions
  • pick the date range of interest
  • check the box on the Sort/Count tab to minimize sampling
  • run your query
  • use the Arrange function to remove the Date column
  • use the Duplicates function to combine duplicate Month of Year values, summing the numbers
  • use the Calculate function to create a new column ‘Bounce Rate’ as Bounces divided by Sessions

Other calculated metrics can be computer in the same way, so when you refresh the data, you will always get the most accurate data available.